Decaflon

<EM>"But even Freud had several grades of consciousness

Neuroscientists should be asking: What phenomenon should I try to explain? Can I make a theory of it?

"Neuroscientists should be asking: What phenomenon should I try to explain? Can I make a theory of it? Then, can I design an experiment to see if one of those theories is better than the others?"</EM>

You are way over my head on this.

What would you expect a pioneer in AI to say anyway? There is so much envy and competition among academic researchers I have no idea what to think half the time.

The idea of picking a phenomena, trying to make a theory of it and then designing an experiment to fit it seems akin to media in which they decide what they are trying to say and only use the points which prove such a thing leaving out all else.

I'll check out your blog post later maybe it will make more sense in its entirety.

Surely neuroscience, to some degree, has explained a lot of characteristic of linguistics/language. I think there is potential for neuroscience to do a lot more.

I admit to being ignorant on the subject of AI.

I want to emphasize something, my take is that he's mainly bashing the idea of neuroscience as a tool to understand consciousness. And he reaches that conclusion by what I can assume is his knowledge on the sheer complexity of the phenomena: We need simulations.

I can say for my part that I went into computer science and AI because I firmly believe it's the best tool to understanding minds. But, of course, like any scientist should, I admit that I could be wrong. Either entirely or on some level. I assume that Minsky does the same; A scientist isn't a reporter or a minister pushing scientifically unsupported beliefs — science is fundamentally concerned with making statements that can be disproven or supported. I honestly think Minsky wouldn't be making these statements if he didn't believe it on a scientific basis.

I agree with you though, that certainly neuroscience is useful for some things — just as most (all) scientific fields. The main issue here is whether it's useful for explaining the mind as a holistic system, to understand what makes it tick (the clock as a whole, the mind) — as opposed to understanding just the parts that are ticking (gears, neurons).

Isn't understanding "parts of "a key to understanding the whole?

Parts of, yes, but not enough. We have to understand how the parts collaborate to produce the whole — there are things to consider such as the emergent effect of interacting components, especially with things as complex as the human brain: 100 billion neurons interacting (if you're interested, the Emergence Wikipedia article is a nice starting point :)). That's basically what Minsky is arguing — that AI and CS are neccessary to understand the overall function of those billions of components.

I think that to fully understand the mind, we (not me personally but the scientist bods) will have to attack the problem from several directions. I think that there are many disciplines which will have something to contribute to the problem.

We can learn about the low level mechanisms, the chemical functions and the foetal development in the brain from neuroscientists, but i don't expect that they can tell us much about the higher levels.

Evolutionary Biology will help us to understand how the brain and mind developed, and what problems it evolved to solve. I think that understanding the intermediate stages will help us to see how the layers of the mind are put together and work together.

We can use ideas from Anthropology to help develop high level theories about how a mind acts as social entity. along with evolutionary biology, this may help us to better understand the historical development of the brain.

Computer Science and AI will help to understand how information can be used and processed, though (personally I feel) not in the classical sense. This will give us some ideas for how the higher level systems interact and produce intelligent systems.

Psychology, attacking the problem from the 'top down' will help us to understand the modularity of the brain, which bits do what, and how they fit together. Though this method will only be able to probe so deep. In the future we may be able to test developing AI theories against psychology experiments.

I agree with craig above; consciousness is a multi-level problem. Perhaps it is similar to physics. There's theoretical and experimental physics, and maths, upon which both of them rest.

Use all the available tools, I say. Be it AI/CS, interpretative dance and painting, psychology.

Though this method will only be able to probe so deep. In the future we may be able to test developing AI theories against psychology experiments.

We're already doing that, actually. Psychology has been extensively used as basis for several prominent AI platforms.

Ultimately, I think the "pieces" excavated from biology, neuroscience, psychology, etc. will be too complex to explain the overall (human) mind by themselves.

We'll have to create a simulation of them — and analyze that simulation in order to gain a deeper understanding of how the various parts interact to produce higher level functions (such as consciousness).

Nonetheless, I agree with you both of course — we need information about these systems from all fields in order to gain a deeper understanding of them.